140.623.01 STATISTICAL METHODS IN PUBLIC HEALTH III

Presents use of generalized linear models for quantitative analysis of data encountered in public health and medicine. Specific models include analysis of variance, analysis of covariance, multiple linear regression, logistic regression, and Cox regression.

Students who successfully master this course will be able to: 1) Use statistical reasoning to formulate public health questions in quantitative terms [1.1 Critique a proposed public health hypothesis to determine its suitability for testing using regression methods and the available data; 1.2 Formulate and correctly interpret a multivariable linear, logistic or survival regression model to estimate a health effect while minimizing confounding and identifying possible effect modification; 1.3 Evaluate the limitations of observational data as evidence for a health effect; 1.4 Appreciate the importance of relying upon many regression models to capture the relationships among a response and predictor in observational studies]; 2) Conduct statistical computations and construct graphical and tabular displays for regression analysis [2.1 Use the statistical analysis package Stata to perform multivariable regression models; 2.2 Document and archive the steps of your statistical analysis by creating a Stata do-file; 2.3 Create and interpret scatter-plots and adjusted variable plots that display the relationships among an outcome and multiple risk factors; 2.4 Create and interpret tables of regression results including unadjusted and adjusted estimates of coefficients with confidence intervals from many models]; 3) Use probability models to describe trends and random variation in public health data [3.1 Distinguish between the underlying probability distributions for modeling continuous, categorical, binary and time-to-event data; 3.2 Recognize the key assumptions underlying a multivariable regression model and judge whether departures in a particular application warrant consultation with a statistical expert]; 4) Use statistical methods for inference in multiple regression to draw valid public health inferences from data [4.1 Conduct a simple linear, logistic or survival regression and correctly interpret the regression coefficients and their confidence interval; 4.2 Conduct a multiple linear, logistic or survival regression and correctly interpret the coefficients and their confidence intervals; 4.3 Examine residuals and adjusted variable plots for inconsistencies between the regression model and patterns in the data and for outliers and high leverage observations; 4.4 Fit and compare different models to explore the association between outcome and predictor variables in an observational study].